Job Market Paper

Can media pluralism be harmful to news quality?

To be presented at the 2021 European Winter Meeting of the Econometric Society and the Asia-Pacific Industrial Organization Conference 2021.

Two stylized facts characterize the Internet: a great diversity of news sources and the proliferation of disinformation. I study a model of information design that connects these observations. I show that competition between news sources with opposite biases reduces information quality when news consumers have limited attention. The reason is the endogenous formation of echo chambers. The standard narrative is that echo chambers arise because news consumers exhibit confirmation bias. I show that even unbiased and rational news consumers devote their limited attention to like-minded news sources in equilibrium. Confirmation bias thus arises endogenously because news sources have no incentive to provide valuable information. I show that the presence of many news sources and the widespread existence of misleading news are concurrent.


Partial compatibility in oligopoly, with Domenico Menicucci (University of Florence), Journal of Economic Behavior & Organization (2021) [Supplementary material] [Old version]

This paper examines the issue of product compatibility in an oligopoly with three multi-product firms. Whereas most of the existing literature focuses on the extreme cases of full compatibility or full incompatibility, we look at asymmetric settings in which some firms make their products compatible with a standard technology and others do not. Our analysis reveals each firm’s individual incentive to adopt the standard, and allows to study a two-stage game in which first each firm chooses its technological regime (compatibility or incompatibility), then price competition occurs given the regime each firm has selected at stage one. When firms are ex ante symmetric, we find that for each firm, compatibility weakly dominates incompatibility. In a setting in which a firm’s products have higher quality than its rivals’ products, individual incentives to make products incompatible emerge, first for the firm with higher quality products, then also for the other firms, as the quality difference increases. This paper sheds lights on markets in which some firms adopt the standard technology but other firms use proprietary systems.

Work in progress

Selective Exposure Reduces Voluntary Contributions: Experimental Evidence from the German Internet Panel with Linnéa Marie Rohde (University of Mannheim)

Can strategic information acquisition harm the provision of public goods? To investigate this question, we conduct an incentivized online experiment on a large and heterogeneous sample from the German population. We implement a one-shot voluntary contribution mechanism where the marginal returns of the public good are uncertain, i.e. whether it is socially efficient to contribute or not. In the main treatment, participants can receive one unit of costless information about the marginal returns from two unreliable information sources with opposing biases. One source is more likely to report low marginal returns, whereas the other is more likely to report high marginal returns. We find that the majority of participants selects the source biased towards claiming low marginal returns, independent of their prior beliefs. As a result, the information treatment significantly reduces contributions and increases free-riding. Moreover, we find that more selfish participants are less likely to acquire information, and if they do so, more likely to select the source that is biased towards claiming high marginal returns than those who have more social preferences. We corroborate our findings by showing that the same participants are also less likely to contribute to actual public goods, namely the environment and COVID-19 containment.

Information Design with Optimal Frame Choice with Yulia Evsyukova (University of Mannheim) and Niccolò Lomys (Toulouse School of Economics)

We study a model of information design where agents have an asymmetric perception of the state space. Whereas the sender (she) perceives all payoff-relevant states of the world, the receiver (he) does not. Persuasion occurs in two stages. The sender first designs the receiver’s optimal frame by expanding or refining his perception of payoff-relevant states. Then, given the chosen frame, the sender designs an optimal information structure. We characterize the sender’s tradeoff between keeping the agent in the dark and expanding/refining his perception of the state space. The optimal frame depends on how the receiver reacts to the discovery of new states or the refinement of his information partition. We characterize optimal persuasion under various standard frameworks to model beliefs under growing awareness: reverse Bayesianism, extended Bayesianism, and partition dependence with sub-additive new beliefs. We also discuss robust approaches to account for the possibility that the sender cannot anticipate how the receiver forms new beliefs after a frame change. Our analysis may shed light on the management of public panic events such as those after the COVID-19 pandemic outbreak.